Conversation
dc91e27 to
b43332c
Compare
|
Would this new package be better places into a different / separate repository? I am not sure that tying it to the release of |
7842991 to
c02dcbe
Compare
It can be moved to another place later. I've added now an script with the tests I made. This is a simple measuring of publish/subscription rate and of latency between publish and subscribe. It would be good to continue the analysis with a performance tool like gperftools. |
c02dcbe to
841e206
Compare
Signed-off-by: ivanpauno <ivanpauno@ekumenlabs.com>
Signed-off-by: ivanpauno <ivanpauno@ekumenlabs.com>
…eads Signed-off-by: ivanpauno <ivanpauno@ekumenlabs.com>
Signed-off-by: ivanpauno <ivanpauno@ekumenlabs.com>
Signed-off-by: ivanpauno <ivanpauno@ekumenlabs.com>
Signed-off-by: ivanpauno <ivanpauno@ekumenlabs.com>
d0cdb4a to
40b84c5
Compare
|
Closing as this is not up to date, and there have been better proposals for measuring performance: |
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
* Move zstd implementation files to new package Signed-off-by: Emerson Knapp <eknapp@amazon.com> * Fix rosbag2_tests builds Signed-off-by: Emerson Knapp <eknapp@amazon.com> * Plugin description needs to specify the new namespace/libraryg Signed-off-by: Emerson Knapp <eknapp@amazon.com> * Add zstd to rosbag2 metapackage Signed-off-by: Emerson Knapp <eknapp@amazon.com> * Remove zstd dependency from rosbag2_compression Signed-off-by: Emerson Knapp <eknapp@amazon.com>
This targets the first task of #634
This package adds three executables:
Script for running some predefined tests, and then do the post-processing and plotting of all of them.
Example usage: ros2 run rclcpp_performance performance_test -d output_dir [--skip]
communication_performance: Which logs the timestamps.
Example usage: ros2 run communication_performance [-d output_dir] [-p publish_period_ms] [-m M][-n N][-unique] [-intra]
posprocess_logging: Plot results.
Example usage: ros2 run posprocess_logging -d output_dir
Where output_dir contains the logged file.
It plots three things:
In all the cases, y axes is in ms and x axes is the result of one of the parts (one publisher, one subscriber, or one publisher-subscriber pair).
My results:
https://docs.google.com/document/d/1YRpWajYvXF_RA7nIH3Qe8RFUyVJwJoxVTSpQFJKN18U/edit?usp=sharing
Old results (841e206):
https://docs.google.com/document/d/11RBqAjcjQ4FBYkwfVtudOaHJSP_wz_TW9AGneo1_ePE/edit?usp=sharing
TODO: